MultiDeepFake: Improving Fake News Detection with a Deep Convolutional Neural Network Using a Multimodal Dataset

Nowadays, the news ecosystem has shifted from traditional print media to social media outlets. It has resulted in the inaccuracy and irrelevancy in updating information by people which is commonly known as fake news. Due to the increasing number of users in social media, fake news is quickly publishing by an individual, and its credibility stands compromised, which brings in a need for effective detection of fake news. Since a large proportion of the population uses social media for updating themselves with news, delivering accurate and altruistic information to them is of utmost importance. Fake news detection has recently garnered much attention from researchers and developers alike. This work proposes to detect fake news using various modalities available, such as text, image, and text in the image in an effective manner using Deep Learning algorithms. In this paper, we propose a deep convolutional neural network for handling diverse multi-domain fake news data. Our proposed model (MultiDeepFake) has obtained more accurate results as compared to the existing state-of-the-art benchmarks. Classification results will motivate the researchers to use our proposed model in future for fake news detection.